Skip to content
GitLab
Explore
Sign in
Tags
Tags give the ability to mark specific points in history as being important
This project is mirrored from
https://github.com/lucidrains/DALLE2-pytorch
. Pull mirroring updated
Sep 19, 2024
.
0.0.101
896f1978
·
remove convnext blocks, they are illsuited for generative work, validated by...
·
May 05, 2022
0.0.100
aec5575d
·
take a bet on resize right, given Katherine is using it
·
May 04, 2022
0.0.99
9773f10d
·
use inference mode whenever possible, cleanup
·
May 04, 2022
0.0.98
86e692d2
·
fix random crop probability
·
May 04, 2022
0.0.97
97b75120
·
allow for last unet in the cascade to be trained on crops, if it is convolution-only
·
May 04, 2022
0.0.96
1992d25c
·
project management
·
May 04, 2022
0.0.95
9359ad2e
·
0.0.95
·
May 04, 2022
0.0.94
58d9b422
·
0.0.94
·
May 04, 2022
0.0.93
70282de2
·
add ability to turn on normformer settings, given @borisdayma reported good...
·
May 02, 2022
0.0.92
11469dc0
·
makes more sense to keep this as True as default, for stability
·
May 02, 2022
0.0.91
0fc6c9cd
·
provide option to l2norm the output of the diffusion prior
·
May 02, 2022
0.0.90
1924c7cc
·
fix issue with mixed precision and gradient clipping
·
May 02, 2022
0.0.89
fc954ee7
·
fix calculation of adaptive weight for vit-vqgan, thanks to @CiaoHe
·
May 02, 2022
0.0.88
ad87bfe2
·
switch to using linear attention for the sparse attention layers within unet,...
·
May 01, 2022
0.0.87
b8cf1e5c
·
more attention
·
May 01, 2022
0.0.86
8b9bbec7
·
project management
·
May 01, 2022
0.0.85
5e421bd5
·
let researchers do the hyperparameter search
·
May 01, 2022
0.0.84
67fcab11
·
add MLP based time conditioning to all convnexts, in addition to cross...
·
May 01, 2022
0.0.82
8260fc93
·
allows one to shortcut sampling at a specific unet number, if one were to be training in stages
·
Apr 30, 2022
0.0.81
ebe01749
·
DecoderTrainer sample method uses the exponentially moving averaged
·
Apr 30, 2022
Prev
1
…
10
11
12
13
14
15
16
17
18
Next